Merge branch 'sb/doc-submodule-sync-recursive'
[git] / compat / precompose_utf8.c
1 /*
2  * Converts filenames from decomposed unicode into precomposed unicode.
3  * Used on MacOS X.
4  */
5
6 #define PRECOMPOSE_UNICODE_C
7
8 #include "cache.h"
9 #include "utf8.h"
10 #include "precompose_utf8.h"
11
12 typedef char *iconv_ibp;
13 static const char *repo_encoding = "UTF-8";
14 static const char *path_encoding = "UTF-8-MAC";
15
16 static size_t has_non_ascii(const char *s, size_t maxlen, size_t *strlen_c)
17 {
18         const uint8_t *ptr = (const uint8_t *)s;
19         size_t strlen_chars = 0;
20         size_t ret = 0;
21
22         if (!ptr || !*ptr)
23                 return 0;
24
25         while (*ptr && maxlen) {
26                 if (*ptr & 0x80)
27                         ret++;
28                 strlen_chars++;
29                 ptr++;
30                 maxlen--;
31         }
32         if (strlen_c)
33                 *strlen_c = strlen_chars;
34
35         return ret;
36 }
37
38
39 void probe_utf8_pathname_composition(void)
40 {
41         struct strbuf path = STRBUF_INIT;
42         static const char *auml_nfc = "\xc3\xa4";
43         static const char *auml_nfd = "\x61\xcc\x88";
44         int output_fd;
45         if (precomposed_unicode != -1)
46                 return; /* We found it defined in the global config, respect it */
47         git_path_buf(&path, "%s", auml_nfc);
48         output_fd = open(path.buf, O_CREAT|O_EXCL|O_RDWR, 0600);
49         if (output_fd >= 0) {
50                 close(output_fd);
51                 git_path_buf(&path, "%s", auml_nfd);
52                 precomposed_unicode = access(path.buf, R_OK) ? 0 : 1;
53                 git_config_set("core.precomposeunicode", precomposed_unicode ? "true" : "false");
54                 git_path_buf(&path, "%s", auml_nfc);
55                 if (unlink(path.buf))
56                         die_errno(_("failed to unlink '%s'"), path.buf);
57         }
58         strbuf_release(&path);
59 }
60
61
62 void precompose_argv(int argc, const char **argv)
63 {
64         int i = 0;
65         const char *oldarg;
66         char *newarg;
67         iconv_t ic_precompose;
68
69         if (precomposed_unicode != 1)
70                 return;
71
72         ic_precompose = iconv_open(repo_encoding, path_encoding);
73         if (ic_precompose == (iconv_t) -1)
74                 return;
75
76         while (i < argc) {
77                 size_t namelen;
78                 oldarg = argv[i];
79                 if (has_non_ascii(oldarg, (size_t)-1, &namelen)) {
80                         newarg = reencode_string_iconv(oldarg, namelen, ic_precompose, NULL);
81                         if (newarg)
82                                 argv[i] = newarg;
83                 }
84                 i++;
85         }
86         iconv_close(ic_precompose);
87 }
88
89
90 PREC_DIR *precompose_utf8_opendir(const char *dirname)
91 {
92         PREC_DIR *prec_dir = xmalloc(sizeof(PREC_DIR));
93         prec_dir->dirent_nfc = xmalloc(sizeof(dirent_prec_psx));
94         prec_dir->dirent_nfc->max_name_len = sizeof(prec_dir->dirent_nfc->d_name);
95
96         prec_dir->dirp = opendir(dirname);
97         if (!prec_dir->dirp) {
98                 free(prec_dir->dirent_nfc);
99                 free(prec_dir);
100                 return NULL;
101         } else {
102                 int ret_errno = errno;
103                 prec_dir->ic_precompose = iconv_open(repo_encoding, path_encoding);
104                 /* if iconv_open() fails, die() in readdir() if needed */
105                 errno = ret_errno;
106         }
107
108         return prec_dir;
109 }
110
111 struct dirent_prec_psx *precompose_utf8_readdir(PREC_DIR *prec_dir)
112 {
113         struct dirent *res;
114         res = readdir(prec_dir->dirp);
115         if (res) {
116                 size_t namelenz = strlen(res->d_name) + 1; /* \0 */
117                 size_t new_maxlen = namelenz;
118
119                 int ret_errno = errno;
120
121                 if (new_maxlen > prec_dir->dirent_nfc->max_name_len) {
122                         size_t new_len = sizeof(dirent_prec_psx) + new_maxlen -
123                                 sizeof(prec_dir->dirent_nfc->d_name);
124
125                         prec_dir->dirent_nfc = xrealloc(prec_dir->dirent_nfc, new_len);
126                         prec_dir->dirent_nfc->max_name_len = new_maxlen;
127                 }
128
129                 prec_dir->dirent_nfc->d_ino  = res->d_ino;
130                 prec_dir->dirent_nfc->d_type = res->d_type;
131
132                 if ((precomposed_unicode == 1) && has_non_ascii(res->d_name, (size_t)-1, NULL)) {
133                         if (prec_dir->ic_precompose == (iconv_t)-1) {
134                                 die("iconv_open(%s,%s) failed, but needed:\n"
135                                                 "    precomposed unicode is not supported.\n"
136                                                 "    If you want to use decomposed unicode, run\n"
137                                                 "    \"git config core.precomposeunicode false\"\n",
138                                                 repo_encoding, path_encoding);
139                         } else {
140                                 iconv_ibp       cp = (iconv_ibp)res->d_name;
141                                 size_t inleft = namelenz;
142                                 char *outpos = &prec_dir->dirent_nfc->d_name[0];
143                                 size_t outsz = prec_dir->dirent_nfc->max_name_len;
144                                 errno = 0;
145                                 iconv(prec_dir->ic_precompose, &cp, &inleft, &outpos, &outsz);
146                                 if (errno || inleft) {
147                                         /*
148                                          * iconv() failed and errno could be E2BIG, EILSEQ, EINVAL, EBADF
149                                          * MacOS X avoids illegal byte sequemces.
150                                          * If they occur on a mounted drive (e.g. NFS) it is not worth to
151                                          * die() for that, but rather let the user see the original name
152                                         */
153                                         namelenz = 0; /* trigger strlcpy */
154                                 }
155                         }
156                 } else
157                         namelenz = 0;
158
159                 if (!namelenz)
160                         strlcpy(prec_dir->dirent_nfc->d_name, res->d_name,
161                                                         prec_dir->dirent_nfc->max_name_len);
162
163                 errno = ret_errno;
164                 return prec_dir->dirent_nfc;
165         }
166         return NULL;
167 }
168
169
170 int precompose_utf8_closedir(PREC_DIR *prec_dir)
171 {
172         int ret_value;
173         int ret_errno;
174         ret_value = closedir(prec_dir->dirp);
175         ret_errno = errno;
176         if (prec_dir->ic_precompose != (iconv_t)-1)
177                 iconv_close(prec_dir->ic_precompose);
178         free(prec_dir->dirent_nfc);
179         free(prec_dir);
180         errno = ret_errno;
181         return ret_value;
182 }